Probabilistic MIMO Symbol Detection with Expectation Consistency Approximate Inference

Javier Céspedes, Pablo M. Olmos, Matilde Sánchez Fernández, Fernando Pérez-Cruz

Abstract

In this paper, we explore low-complexity probabilistic algorithms for soft symbol detection in high-dimensional multiple-input multiple-output (MIMO) systems. We present a novel algorithm based on the Expectation Consistency (EC) framework, which describes the approximate inference problem as an optimization over a non-convex function. EC generalizes algorithms such as Belief Propagation and Expectation Propagation. For the MIMO symbol detection problem, we discuss feasible methods to find stationary points of the EC function and explore their tradeoffs between accuracy and speed of convergence. The accuracy is studied, first in terms of input-output mutual information and show that the proposed EC MIMO detector greatly improves state-of-the-art methods, with a complexity order cubic in the number of transmitting antennas. Second, these gains are corroborated by combining the probabilistic output of the EC detector with a low-density parity-check (LDPC) channel code.

Introduction

In Multiple-Input Multiple-Output (MIMO) systems, multiple antennas transmit and receive data simultaneously to boost spectral efficiency. A critical challenge at the receiver is optimal symbol detection: accurately recovering the transmitted symbols despite significant inter-antenna interference.

While the optimal solution provides the best performance, it is computationally prohibitive, scaling exponentially with the number of antennas and the modulation order (e.g., 256-QAM). Existing approximations like Minimum Mean-Squared Error (MMSE) are efficient but suffer from significant performance loss. Advanced methods like Sphere Decoding or Markov Chain Monte Carlo (MCMC) often fail to scale to "massive MIMO" regimes or are too slow for real-time implementation. This paper addresses this gap by proposing a near-optimal "soft" (probabilistic) detector that maintains a manageable computational footprint.

Methods and Framework

The core innovation of this work is the application of the Expectation Consistency (EC) framework to the MIMO detection problem. EC treats inference as an optimization task over a "free energy" function, generalizing well-known techniques like Belief Propagation (BP) and Expectation Propagation (EP).

The EC MIMO Detector

The authors propose a Single-Loop (SL) update rule that iteratively refines the estimate of transmitted symbols. To ensure numerical stability, especially at low signal-to-noise ratios (SNR), they introduce a "progressive variance limit" heuristic. This prevents the algorithm from becoming prematurely confident in its estimates, thereby avoiding poor local minima.

Computational Complexity

A primary advantage of the EC detector is its efficiency. As shown in the complexity analysis, the proposed algorithm maintains a complexity order of $O(m^3)$ per iteration, where $m$ is the number of transmit antennas. This cubic scaling is comparable to standard linear detectors but offers significantly better performance.

Table 1: Complexity order comparison of various MIMO detectors. The EC-based methods maintain polynomial complexity relative to the number of antennas.

Experimental Setup

The method was evaluated using standard simulated channel models:

  • Channel Model: Memoryless flat-fading complex MIMO channels with i.i.d. Rayleigh fading.
  • Modulations: QPSK, 16-QAM, 64-QAM, and 256-QAM.
  • System Sizes: Configurations ranging from $5 \times 5$ (for comparison with optimal detectors) to large $32 \times 32$ arrays.
  • Channel Coding: Integration with optimized Low-Density Parity-Check (LDPC) codes to evaluate real-world decoding performance.

Key Results

The EC detector demonstrates remarkable robustness and accuracy across various scenarios:

  1. Mutual Information: The EC detector significantly reduces the gap to theoretical channel capacity. As illustrated in the performance plots, it outperforms existing state-of-the-art methods like Gaussian Tree Approximation (GTA) and CHEMP.
  2. Near-Optimal Performance: In smaller systems where the optimal solution is computable, the EC detector's performance curve almost perfectly overlaps with the optimal detector.
  3. Bit Error Rate (BER) Gains: When combined with LDPC codes in a $5 \times 5$ system, the EC detector is within 0.4 dB of the optimal detector, whereas the GTA method lags by more than 1.5 dB.
  4. Scalability: In $32 \times 32$ systems with 16-QAM, the EC detector achieves a stable gain of 2.5 dB over the GTA method.
Figure 1: Average mutual information per dimension vs. SNR for various detectors. The EC detector (implied by the performance gains discussed) significantly outperforms MMSE and GTA, approaching the optimal detector's performance.

Conclusion and Future Work

By framing symbol detection as a free-energy optimization, the EC framework provides a powerful tool for high-dimensional wireless communication. It offers a superior trade-off between complexity and accuracy compared to traditional message-passing algorithms.

Future research directions include further reducing complexity for "extremely" massive MIMO systems (hundreds of antennas) and exploring Turbo Receivers, where the EC detector could be integrated into an iterative loop with the channel decoder for even greater performance gains.

Summary updated

Abstract

In this paper we explore low-complexity probabilistic algorithms for soft symbol detection in high-dimensional multiple-input multiple-output (MIMO) systems. We present a novel algorithm based on the Expectation Consistency (EC) framework, which describes the approximate inference problem as an optimization over a non-convex function. EC generalizes algorithms such as Belief Propagation and Expectation Propagation. For the MIMO symbol detection problem, we discuss feasible methods to find stationary points of the EC function and explore their tradeoffs between accuracy and speed of convergence. The accuracy is studied, first in terms of input-output mutual information and show that the proposed EC MIMO detector greatly improves state-of-the-art methods, with a complexity order cubic in the number of transmitting antennas. Second, these gains are corroborated by combining the probabilistic output of the EC detector with a low-density parity-check (LDPC) channel code.
BibTeX
@article{cespedes2018probabilistic,
  author = {Javier Céspedes and Pablo M. Olmos and Matilde Sánchez Fernández and Fernando Pérez-Cruz},
  title = {{Probabilistic MIMO Symbol Detection with Expectation Consistency Approximate Inference}},
  journal = {IEEE Trans. Vehicular Technology},
  year = {2018},
  url = {https://openreview.net/forum?id=WwtZX04sV2},
}
Figure